Skip to content

Chapter 17: Production Geospatial Architecture

Production geospatial architecture

Production geospatial architecture is the discipline of making spatial systems reliable, observable, scalable, affordable, secure, and maintainable.

Learning Goals

  • Design reference architectures for geospatial products.
  • Choose storage, compute, cache, and API patterns.
  • Build observability and testing into spatial platforms.
  • Plan migrations from legacy GIS to modern systems.

Theory

Production architecture is about tradeoffs. Raw data may live in object storage. Queryable operational data may live in PostGIS. Large analytical data may live in GeoParquet or a warehouse. Fast display may require tiles. Expensive analysis may require precomputation. Realtime systems may require streams and state stores.

Good architecture separates source of truth, derived products, caches, and user-facing views.

Math

Relevant math includes spatial partitioning, tile pyramids, cache cardinality, query complexity, graph complexity, raster chunking, throughput, latency percentiles, and cost modeling.

Equation companion: Math and Algorithms Reference

Tools of the Trade

  • PostGIS, object storage, GeoParquet, COG, Zarr.
  • Vector tile servers, CDN caches, map rendering services.
  • Containers, Kubernetes, serverless functions, queues.
  • Terraform, CI/CD, observability stacks.
  • Data catalogs and lineage systems.

Examples of Real-World Solutions

  • A national data portal serves public layers through cached tiles and APIs.
  • A wildfire dashboard precomputes hazard summaries and streams incident updates.
  • A city migrates shapefile-based workflows into PostGIS and automated publishing.
  • A climate analytics service stores rasters in cloud-native chunks and computes summaries on demand.

Working Practice Examples

  1. Draw an architecture for a parcel viewer with editing and public read-only tiles.
  2. Add observability metrics for a tile service.
  3. Estimate storage and query cost for a 10 TB raster archive.
  4. Plan a migration from desktop-only GIS to a shared spatial database.

Common Failure Modes

  • One database trying to do every workload.
  • No cache invalidation strategy.
  • No rebuild path for derived products.
  • Poor observability for data quality.
  • Architecture that depends on manual GIS exports.

Works Cited

Kleppmann, Martin. Designing Data-Intensive Applications. O'Reilly Media, 2017.

"Cloud Optimized GeoTIFF." Cloud Optimized GeoTIFF, https://www.cogeo.org/. Accessed 9 May 2026.

"GeoParquet Specification." GeoParquet, https://geoparquet.org/. Accessed 9 May 2026.

"PostGIS Manual." PostGIS, https://postgis.net/docs/. Accessed 9 May 2026.

GeoInformatica Consulting logo